Rifkin's Pesky Idea

by Bernard D. Sherman

Reprinted from Early Music America, Summer 1999, p. 48; the version here is slightly expanded and revised.

My friend Carole loves Latin-texted sacred music, but she didn't like Bach's B Minor Mass. The recordings in her collection, she said, didn't convey what she calls the "true feeling of the mass." Out of curiosity, I sent her Joshua Rifkin's 1982 recording, with one singer per part. Was I doing her a favor? Not to judge by Ted Libbey's recent review of B Minor Mass recordings in Early Music America (Spring, 1999). Rifkin's recording, he says, is one "to be aware of, but not enamored of."

Mr. Libbey's review displayed his typically high share of sensitive insight. But when he dismissed not only the recording, but also Rifkin's idea that Bach typically used only one singer on each choral line, Libbey must have known that he was stepping onto a hornet's nest. The controversy is as lively as ever.

That it still draws any breath at all is notable: wacky theories of Bach performance, such as the curved "Bach Bow," have had a way of disappearing into the graveyard of interpretive fads. Why doesn't Rifkin's pesky idea follow suit? I think there's good reason.

Consider Libbey's main objection, one that has been raised repeatedly since Rifkin first presented his ideas in 1981: What about Bach's memo to the Leipzig city council, the "Entwurff"? This 1730 letter reads like a call for three or, preferably, four singers per part. It seems to be a straightforward statement of the composer's "ideal," and that's how it was always read by Bach scholars - until Rifkin addressed it.

Rifkin, Andrew Parrott, and John Butt have argued that when read with philological care, this document tells us nothing clear about Bach's choral preferences. They point out that the passage about vocal forces refers not to Bach's four-voiced cantatas, but to simpler motets by older composers - motets typically for eight voices, not four. Moreover, they say, the numbers Bach gives refer not to the "starting lineup" (to use Rifkin's favorite baseball analogy) but to the "roster" of team members necessary to staff an entire church-year's worth of singing in more than one church. The text can even be interpreted as supporting Rifkin's views. It is not strong evidence for Rifkin, of course, nor does he claim it is. But he and others have shown that is is in no way evidence against him.

Similar weaknesses undermine many other objections to Rifkin's idea. And virtually none of the objections have addressed the positive evidence, drawn principally from specific parts used by Bach's singers, on which Rifkin has based his arguments (see below). One begins to see why the hornets haven't settled down yet.

Mr. Libbey also raises a purely pragmatic objection, which is heard often enough: he accuses Rifkin of putting "theory before practicality, something Bach would never do." Libbey attributes to to Robert Shaw the claim that in live performance three trumpets would drown out the one-per-part singers - unless, Shaw added, the singers are aided by studio microphones.

But on this point the great, lamented choral director was mistaken. Both Rifkin and Parrott insist that the balances in their recordings do not reflect studio trickery. Their assertions gain objective support from a tape (provided to me by the Regensburg Festival) of a 1997 concert performance of the B Minor Mass by Rifkin. The recorded sound is reminiscent of the 1940s, since the tape was made by an amateur sitting in the audience, using a single microphone and a home machine. Yet the balances are excellent.

That performance and many others have shown that Rifkin's approach works perfectly well. Its practicability doesn't prove that Bach performed this way, but it does undermine the objection that Rifkin put "theory before practice."

Mr. Libbey argues, finally, that a reinforced chorus had theological meaning for Bach: it symbolizes that the faithful are a group. Yet Eric Van Tassel has appealed to Lutheran theology to argue in favor of one-per-part scoring; he speculates that having one singer per part could symbolize "a community in which each believer is an individual answerable to God." John Butt* finds theological meanings in one-per-part scoring in the Passions; for example, when the singer of Christ's part also sings in the choruses of his tormentors, it symbolizes that "we are all to be held responsible for Christ's death, however Christ-like any of us may seem." You needn't take sides here to recognize that Mr. Libbey's theological argument is no more conclusive than the objections mentioned earlier.

I myself once dismissed Rifkin's idea as being the equivalent of a JFK conspiracy theory, but after more investigation I came to regard it as the most plausible reading of the evidence. I am not alone. Several leading Bach scholars - Butt, David Schulenberg, Daniel Melamed, and Jeanne Swack, for example - have begun to support it in print, and other younger scholars support it in conversation. At the same time, a number of respected artists, including Parrott, Butt, Konrad Junghänel, Sigiswald Kuijken, Paul McCreesh, Marc Minkowski, and Jeffrey Thomas have begun to perform Bach with one singer per part. The growing acceptance by artists of such caliber does not, again, prove that the idea is right, but it does suggest that it is plausible - and that its artistic results can convince some first-rank musical ears.

They don't convince all such ears, of course, including those of Mr. Libbey. I am certainly not a more subtle or judicious critic than he is; but I can testify that one's ears can adjust to what he calls "Rifkinite" Bach. My friend Carole didn't even need time to adjust. "What a difference!" she wrote; indeed, she "had always hated the opening Kyrie until I heard the tape you sent." Subjective reactions, pro or con, tell us nothing about the historical truth of Rifkin's thesis; but Carole's response and mine suggest that one-per-part Bach is not just a scholarly exercise.

Rifkin's pesky idea has by no means triumphed. Some very smart people, including Ted Libbey (and several distinguished Bach scholars), still oppose it. But it is a stronger thesis than many of its critics seem to realize, in its ability both to answer to the evidence and to translate into practice. I expect that the hornets will be buzzing for a long time to come.


Click here for Joshua Rifkin's response, reprinted from Early Music America, Fall 1999, p. 48

FOR FURTHER READING: At last, a book is available on the one-per-part issue. Andrew Parrott's The Essential Bach Choir makes a strong, carefully constructed case for Rifkin's idea. Not only is the book thorough and well-argued; it is also convenient, in that it collects mountains of evidence that previously had been scattered over dozens of sources, as well as providing new evidence. It is published in the UK by Boydell & Brewer and in the US by University of Rochester Press.

I should add that other objections have been raised to the one-per-part idea besides those discussed in my brief op-ed. Several critics, for example, have pointed to other locales in Germany where, they believe, larger choruses were the norm; these choruses, it is said, represent Bach's ideal. But such arguments have proven no more durable than the other objections I discussed. The matter is discussed in detail in Parrott's book.

1. A brief summary of Rifkin's idea. Here is a revised version of a footnote from my book Inside Early Music :

Rifkin’s logic can be expressed in a syllogism. The major premise is that Bach provided his singers with written parts meant for use by a single performer only. The minor premise is that in all but a few of Bach’s surviving choral works only one part was ever written out for each of the voice ranges. If both premises are accepted, the conclusion is obvious: Bach used a solo quartet to sing the choruses.

The minor premise, that Bach usually prepared only one part for each range, is standard Bach scholars. The major premise, however, is revolutionary. Scholars have long believed that Bach’s singers shared their parts, typically in groups of three – a principal singer, who sang everything in the part from beginning to end, and two supporting “ripienists,” who joined in only for the choruses and chorales. But according to Rifkin, there is no evidence of the ripienists sharing the parts, either in documents or in the parts themselves. And, as we'll see, that there is positive evidence to the contrary.

What is that evidence against part-sharing? Let me repeat that all commentators agree that a typical Bach voice part included every movement in that vocal range - both solo and choral - in the entire work. If the ripienists read from such a part, they would have to know when to join in the singing and when to stay silent (again, the principal singer, all agree, would have been singing from beginning to end). Yet the parts consistently fail to provide cues of any sort for when ripieno singers should enter and exit - even in complex situations that seem to demand them. Bach did take the trouble to notate detailed performance cues into these parts (about articulation, dynamics, ornamentation, and even tempo). Thus it seems unlikely that he didn't have time to notate something so basic as ripieno entries. Nor is it easy to understand why he would feel he could leave that particular detail for verbal instruction in rehearsal.

Rifkin points out that even those few works for which Bach did provide special extra ripieno parts (such as the last revision of the St. John Passion) logically support the major premise. In these works there is no question that the principal parts are meant just for single voices.

As for the substantial number of parts that show positive signs of having been intended for use by one singer only, Rifkin points to several that bear character designations (“Anima,” “Jesus,” "Evangelista," “Phoebus”), and to some that indicate solo singers in the list of forces on the wrapper in which Bach kept the parts (the vocal scoring for BWV 56 reads “S. A. T. &Basso Con[certato]”). John Butt, in his book on the Passions, Bach's Dialogue with Modernity, gives another example of this principle at work in the parts written out for Jesus and the Evangelist in the St. Matthew Passion: on the one hand, the parts have markings telling the solo singers to be silent ("tacet") during many choral sections, but on the other they have no markings (like "tutti") indicating when the supposed choral onlookers would have entered during choruses.

Butt notes another significant piece of positive evidence for Rifkin's theory in the St. Matthew Passion: when the duet "So ist mein Jesus nun gefangen" moves directly into the fearsome double chorus"Sind Blitze, sind Donner," Bach (who was running out of staves on the page) saved space by writing the words "due Tenore" over the tenor's line, after writing "Basso 1 Chori/ Basso 2 Chori Concord." over the bass entry, and "Tutti li bassi in unisono" over the bass's continuo line. The notation in the tenor line makes sense if there was just one tenor in each of the two choirs.

Back to text

*. John Butt, in "Bach's vocal scoring: What can it mean?" Early Music, February, 1998, pp. 99-107.Back to text

. Growing Academic Support: John Butt argues in favor of Rifkin's idea (and contributes more evidence for it) in his "Bach's vocal scoring." Previously, in his Bach: Mass in B Minor (p. 40), he wrote, “although [Rifkin’s] view continues to be opposed by some of the most important figures in Bach research, there have been no convincing arguments, based on meticulous source-study, actually to prove him wrong.” George Stauffer disputes this in the Journal of Musicological Research 12 (1993), pp. 257—72; Rifkin replies in the same journal, 14, (1995), pp. 223-34. Jeanne Swack has presented research showing that Telemann used one-per-part scoring in his cantatas. Kerala Snyder, in her 1987 book Buxtehude, shows that this composer - whose performances so captivated the young Bach - normally intended his four-voice choral compositions for soloists. Finally, David Schulenberg's Music of the Baroque (Oxford, 2001) - the leading textbook on the period - endorses Rifkin's viewpoint. He writes: "Although the exact makeup of Bach's vocal forces has been a matter of debate, it appears increasingly likely that most of Bach's vocal works were composed for a 'chorus' comprising a single singer on each part. Orchestral parts, too, were rarely doubled, except for the violin and continuo lines. Thus what many modern listeners have come to regard as massive choral movements for large choir and orchestra are in fact examples of chamber music for vocal soloists and a small instrumental ensemble." (p. 192). Whether or not you agree, when the standard textbook presents an idea as "increasingly likely" that idea can no longer be regarded as being on the fringe. Back to text

Return to Bernard D. Sherman's Web site